The proposed research will continue our exploration of the critiquing approach to bringing artificial intelligence (AI) based advice to the practicing physician. To date, our research has primarily involved the implementation of several prototype critiquing systems in different medical domains. The proposed research will focus on one of these systems, HT-ATTENDING, which critiques the pharmacologic management of essential hypertension. 1. A Controlled Experiment A central component of the proposed project will involve a controlled experiment to help evaluate HT- ATTENDING in Yale's Primary Care Center. 2. An Experimental Period of Clinical Use Following the formal evaluation study, there will be a period of clinical use of HT- ATTENDING at the Primary Care Center. During this period, we will test various augmentations and modifications of the system. 3. Refining, Augmenting, and Updating HT-ATTENDING Throughout the proposed project, we will continually refine HT-ATTENDING and maintain the currency of its knowledge base. 4. Tools for Knowledge Inspection, Verification, and Updating We will continue to develop and refine computer-based tools to assist in the inspection, verification, and updating of HT-ATTENDING's knowledge base. 5. An External Validation of HT-ATTENDING's Knowledge He will perform an "external" validation of HT-ATTENDING's knowledge by submitting 25 critiques of varying complexity to outside ypertension experts.